Lipids of Wheat Flour. I. Characterization of Galactosylglycerol

-1 benzene extract of bleached wheat flour was fractionated by Craig distribution between n-heptane and 95c"o methanol into triglyceride, steroid, lip...
3 downloads 0 Views 564KB Size
A u g . 5 , 19.56

GALSCTOSYLGLYCERO1,S F R O M WHEAT [CONTRIBUTION FROM THE NOYES CHEMICAL

Lipids of Wheat Flour. I.

FLOUR

3735

LABORATORY, UNIVERSITY OF ILLINOIS]

Characterization of Galactosylglycerol Components'

BY H. E. CARTER, R. H. MCCLUER~ AND E. D. SLIFER RECEIVED MARCH 7, 1956 -1benzene extract of bleached wheat flour was fractionated by Craig distribution between n-heptane and 95c"o methanol into triglyceride, steroid, lipoprotein and lipocarbohydrate fractions The lipocarbohydrate was further fractionated from acetone into a more soluble and a less soluble component In a study of these two materials the carbohydrate moiety of the more soluble fraction was identified as P-D-galactopyranosyl-1-glyceroland the carbohydrate moiety of the less soluble fraction as a-D-ga~actopyranos~~-~,6-~-D-ga~actopyranosy~-l-g~ycero~

A study of the benzene-extractable lipids of bleached wheat flour has demonstrated the presence of a t least two galactosylglyceride compounds. The purpose of this paper is to report the characterization of the carbohydrate moiety of these lipids. In an initial study the benzene extract was separated by counter-current distribution between n-heptane and 95% methanol into four distinct fractions. The four fractions consisted of a steroid which precipitated spontaneously from the nheptane, a lipoprotein which was precipitated from both phases with acetone, a triglyceride fraction from the n-heptane phase and a fraction containing 20% carbohydrate from the methanol phase. The only material to be described in this paper is the lipocarbohydrate fraction, the remaining materials will be the subject of future studies. The lipocarbohydrate fraction was soluble in warm acetone but on cooling could be separated into acetone-insoluble and soluble materials, designated fractions I and 11, respectively. To date, attempts to obtain completely homogeneous substances by this procedure have been unsuccessful. Analytical data on these two fractions are presented in Table I. TABLE I LIPOCARBOHYDRATE FRACTIONS I AND I1 Carbohydrate,

Fatty acid,

Nitrogen,

29 12

58 56

0.59 0.18

%

I I1

7a

%

Phosphorus,

% 0.26 0.5

Chlorine,

%

4.6 3.6

Fractions I and I1 were subjected to alkaline hydrolysis. The hydrolysates were acidified and fatty acids were extracted with petroleum ether. The aqueous solutions containing the carbohydrate moieties were deionized and finally chromatographed on carbon-Celite columns. Elution with dilute aqueous ethanol yielded two distinct carbohydrate fractions, A and B. Fraction I yielded 10% of A and 90yo of B, whereas from fraction 11, A and B were obtained in approximately equal quantities. Both A and B were chromatographically homogeneous in three solvent systems and were non-reducing before acid hydrolysis. Upon acid hydrolysis A and B yielded only galactose and (1) This investigation was supported by a research grant from Procter and Gamble Co. T h e authors wish t o express their appreciation for t h e grant and for a generous supply of wheat flour lipids. Part of the material in this paper was taken from the thesis submitted by E. D. Slifer to t h e Graduate College of t h e University of Illinois in partial fulfillment of the requirement for t h e degree of Doctor of Philosophy in Chemistry. (2) Post-doctorate Research Associate in Chemistry, 1955-1956,

glycerol as determined by paper chromatography. Crystallization of fraction A was accomplished from absolute methanol. Fraction B was crystallized from an ethanol-water mixture. Data shown in Table I1 indicate that fraction A is a monogalactosylglycerol and fraction B a digalactosylglycerol. TABLE I1 ANALYSES O F COMPOUNDS Monogalactosyiglycerol Fraction TheoA retical

Galactose, yo Glycerol, yo M.p., " C . [a] 27D

71 35.3 139-140 +3.77O

70.9 35.8

. ... ..

A

AND

B

Digalactosylglycerol Fraction Theon retical

84 22.8 182-184 +86.4'

86.4 22.9 ..

.

, . .

ildditional evidence for the presence of galactose as the only carbohydrate component of A and B was obtained by preparing the a-methylphenylhydrazone of galactose from acid hydrolysates of these substances. The over-all yield of this derivative from -4 and B was 85 and goyo,respectively. To further elucidate the structure of these compounds, a periodate oxidation study was made. The monogalactosylglycerol (A) consumed 3 moles of periodate and produced 1 mole of formic acid and 1 mole of formaldehyde. These data are consistent only with a galactopyranosyl-1-glycerol structure. The digalactosylglycerol (B) consumed 5 moles of periodate, yielding 2 moles of formic acid and 1 mole of formaldehyde. Thus a galactopyranosyl-1,6-galactopyranosyl-l-glycerolstructure was indicated. Studies of the enzymatic hydrolysis of A and B with specific a- and @-galactosidasesconfirmed the periodate results. When the di-galactosylglycerol (B) was incubated with yeast a-galactosidase, onehalf of the potential reducing power was liberated. Chromatography of the hydrolysis products on a carbon-Celite column resulted in the isolation of galactose and a monogalactosylglycerol which was identical with ,4. Treatment of either preparation of monogalactosylglycerol with a-galactosidase resulted in no demonstrable hydrolysis. An E. coli 8-galactosidase preparation gave no detectable hydrolysis of the digalactosylglycerol but cleaved the monogalactosylglycerol a t a moderate rate. These results establish a /?-linkage of galactose to glycerol in both compounds and an a-linkage of the second galactose residue in the digalactosylglycerol. The specific rotations of A and B supply supplementary evidence for the nature of the glycosidic linkages. /?-Galactosides usually have a very low

H.E.CARTER, R.H. MCCLUERA N D E. D. SLIFER

37313

Vol. 7s

specific rotation (P-methylgalactoside, 0.03)whereas it was necessary t o centrifuge the emulsified phase for 15 minutes a t 2000 r.p.m. The clear upper phase W R S demost a-galactosides have specific rotations4 of 125 canted off and added back to number l heptane. The to 200'. Thus the low rotation (f3.77') of the lower phase was siphoned off and added to the number 2 monogalactosylglycerol is characteristic of a p- heptane funnel. A white interfacial material remained. galactoside while the value of +86.4" for the This was resuspended in 20 ml. of n-heptane and filtered by suction on a sintered glass funnel. The residue was washed digalactosylglycerol is about that to be expected with another 10-ml. portion of n-heptane and the heptane from a compound containing one a- and one /3- wash was added to number 1 heptane. Number 9 methanol galactosidic linkage. WV:LS then equilibrated with number 2 heptane and the process The following structures are therefore assigned was continued until each of the methanol tubes had been equilibrated in turn with each of the n-heptane tubes. I n to the two carbohydrates. terfacial material resulted only in the first equilibration. CH,OH

CHIOC

I

CHOH

H OH @-D-galactopyranosyl- 1-glycerol ( A ) CHIOH

,I-o.

CHIOH

I

CHOH

H OH a-D-galactopyranosyl-1,6-B-D-galactopyranosyl- 1-glycerol (B)

Putman and Hassidj have reported the isolation of cu-~-galactopyranosyl-2-glycerol from the red marine alga Irideae laminarioides. Dr. Hassid kindly provided us with a small amount of this material for comparative studies. The alga product was found to be distinct from our monogalactosylglycerol by paper chromatography. optical rotation, melting point, and behavior toward enzymatic hydrolysis. The isolation of two different monogalactosylglycerols from plant sources provides an interesting problem in biochemical relationships. Experimental Fractionation of Wheat Flour Lipids.-In the first fractionation a nine transfer, double withdrawal Craig distribution was carried out in the usual manner. The upper phase consisted of four 500-1111. portions of n-heptane which had been pre-equilibrated with 95% methanol; the lower phase of five 500-ml. portions of 95y0 methanol pre-equilibrated with n-heptane. The distributions were carried out in 2-1. separatory funnels (numbered 1-9). At the end of the distribution, funnels 1-4 contained the upper phase and funnels 5-9 the lower phase, with funnel 9 representing the first methanol phase to come in contact with number 1 n-heptane. Thirty-eight grams of benzene-extracted wheat flour lipids was dissolved in number 1 heptane and this solution was equilibrated with number 9 methanol. I n this first equilibration a n emulsion resulted in the lower phase and (3) J. K. Dale and C. S. Hudson, THISJOURNAL, 62, 2534 (1930). (4) F. S. Bates, "Polarimetry, Saccharimetry, and t h e Sugars," United States Government Printing Office, Washington, 1942, p. 718. (5) E. TV. Putman and W. 2. Hassid, THISJ O U R N A L , 76, 2221 (1954).

On completion of the distribution the solvent was removed from each of the 9 fractions by evaporation at room temperature. As the n-heptane was removed from the number 1 tube a steroid crystallized out and was removed by filtration on a sintered glass funnel. These nine fractions were then treated with acetone which resulted in a further fractionation of each into a n acetone-soluble and a n acetoneinsoluble fraction. An analysis of each of the eighteen fractions as well as of the interfacial material for carbohydrate! phosphorus and nitrogen suggested t h a t at least four distinct fractions were present. All of the acetone-insoluble material as well as the interfacial material contained 2% phosphorus, 6% nitrogen and from 5 t o 7% carbohydrate. T h e acetonesoluble material in the n-heptane phase contained less than 1% carbohydrate, and less than 0.2% each of phosphorus and nitrogen while the acetone-soluble material from the methanol phase contained less t h a n 0.5% each of phosphorus and nitrogen but contained approximately 20% carbohydrate. Since over 80% of the material was present in funnels 1, 8 and 9 in the nine transfer distribution, a second distribution was carried out with only three transfers. One hundred grams of benzene extract was dissolved in 2 1. of n-heptane, pre-equilibrated with 95% methanol, in a 4-1. separatory funnel and extracted twice with 1 1. of pre-equilibrated methanol. The first methanol extract was referred t o as tube number 3 and the second methanol extract as tube number 2. Tube 1 (heptane phase) contained 62.8 g. of lipid; tube number 2, 10 g.; tube number 3, 21.3 g. I n addition 2.9 g. of interfacial material was obtained in the first distribution. The material from tube number 2 was extracted twice with 150-ml. portions of hot acetone. The residue, consisting of 1.0 g. of lipoprotein, was set aside. The acetone solutions were concentrated to one-half volume and cooled to 0' giving 4.0 g. of white solid which melted a t room temperature and resolidified t o a white paraffin-like wax (fraction I ) . The acetone filtrate on removal of the solvent gave 4.0 g. of viscous brown sirup (fraction 11). The material from tube number 3 was fractionated in the same way, giving 3.0 g. of lipoprotein, 9.4 g. of fraction I and 7.0 g. of fraction 11. Isolation of Carbohydrate Moieties of Fraction 1.-One gram of material in 30 ml. of 0.1 N sodium hydroxide was hydrolyzed for 5 hours in a boiling water-bath. Most of the material had dissolved in the first hour. After hydrolysis the reaction mixture was cooled and 0.1 N hydrochloric acid was added t o a phenolphthalein end-point. These data indicated t h a t 3.42 meq. of alkali was required to hydrolyze 1 g. of the intact lipid. The alkaline hydrolysate was acidified t o p H 1.0 with concentrated hydrochloric acid and the fatty acids liberated were extracted with petroleum ether. Removal of the solvent gave 0.578 g. of residue per g. of starting material. If the mean molecular weight of the fatty acids is calculated on the basis of the alkali required for the hydrolysis, a value of 165-170 is obtained. However, when the extracted fatty acids were titrated with alcoholic potassium hydroxide only 1.61 meq. of alkali was required. This discrepancy is undoubtedly due in part to the presence of chlorine in the intact material. The mean molecular weight of the fatty acids based on this latter figure is 359. The nature of these fatty acids is under investigation. After removal of the fatty acids by extraction with petroleum ether the hydrolysate was deionized by passing the aqueous solution over Amberlite resins I R 120 in the hydrogen phase and I R 400 in the hydroxyl phase. (In subsequent experiments it was necessary occasionally t o repeat the treatment with a freshly charged column of I R 400 or

GALACTOSYLGLYCEROLS FROM WHEATFLOUR

Aug. 5 , 1933

3737

TABLE I11 PAPER

Solventa

e

A

CHROMATOGRAPHY O F A I?! and Values

A hyd.d

B

AND

B'

B hyd.d

@-Galactosyl2-glycerol

Galactose

Glycero1

Pyr.:but.:water* 0.46 0.46, 0.67 0.23 0.46,0.67 0.51 0.46 0.67 sym-Collidinee 0.95 1.0, 1.58 0.40 1.0, 1.58 0.98 1.0 1.58 Compounds detected with 5% ammoniacal silver nitrate. Pyridine:%-butyl alcoho1:water (3:2: 1.5); Rr values. sym-Collidine saturated with water; R ~ ~values. i ~ ~Hydrolyzed t ~ ~in 1 ~ N hydrochloric acid for 40 min. a t 100".

I R 120 t o obtain a neutral solution.) The Molisch and anthrone positive components washed on through the resin columns with little or no holdup. T h e neutral aqueous solution was lyophilized giving a hygroscopic solid residue. T h e procedure of Whistler and Dursoe was used to isolate the carbohydrate components. This procedure utilizes a carbon-Celite column composed of Darco G-60 and Celite 545 in a 1: 1 ratio as the adsorbent. T h e carbohydrate residue was redissolved in 10 mi. of distilled water. The solution was placed on a 2 X 8 cm. column and elution with water begun. Three hundred ml. of aqueous wash followed by 200 ml. of a 2 % ethanol solution yielded no Molisch positive material. Then washing with 5 % ethanol was begun, 5-ml. fractions being collected. The carbohydrate content of the fractions was followed by a quantitative anthrone procedure. Two distinct peaks were obtained. The first fraction (A) came off the column in tubes 13 and 20 and the second (B) in tubes 42 t o 80. Elution with 95% ethanol did not result in any more Molisch positive material. One gram of fraction I yielded 25-30 mg. of A and 225-270 mg. of B. Isolation of Carbohydrate Moieties of Fraction 11.-This fraction was hydrolyzed with alkali, deionized and the carbohydrate components isolated in exactly the same manner a s described above. However, in the elution of the material from the carbon-Celite column, the first fraction was obtained with an aqueous wash while the second fraction required a 2.5% ethanol solution for elution. These two fractions were obtained in about equal quantities. The first fraction t o be eluted was shown by the paper chromatographic procedure described below t o be identical with the fraction A previously obtained while the second was identical with fraction B. Crystallization of A and B.-A portion of fraction A was dissolved in a minimum quantity of absolute methanol and stored a t 4'. After about two months a significant amount of crystalline material had formed. These crystals were collected and used t o seed any subsequent crystallizations. This material could be recrystallized from absolute methanolethyl ether mixture although much smaller crystals resulted. Fraction B was crystallized from an ethanol-water mixture. A portion of this material was dissolved in a minimum quantity of water and absolute ethanol was added until the solution became turbid. T h e solution was warmed on the steam-cone until it became clear and then allowed t o stand a t room temperature overnight. Clusters of white needles collected on the walls of the container. Characterization of A and B.-Paper chromatograms containing up t o 150 pg. of fraction A or B gave only one spot when they were sprayed with a 5% ammoniacal silver nitrate spray reagent, dried, and heated in a n oven a t 110' for 5 minutes. The materials gave reddish-brown spots typical of non-reducing substances. After the materials had been hydrolyzed for 40 minutes a t 100" with 1 N hydrochloric acid, and subjected to chromatography, two spots appeared with the silver nitrate spray. One corresponded t o a glycerol standard and the other t o the galactose standard. When the papers were sprayed with aniline hydrogen phthalate only one spot corresponding to the galactose standard appeared. Paper chromatographic data are summarized in Table 111. Glycerol analyses were carried out according t o Blix.7 Both A and B gave mucic acid on treatment with nitric acid. The galactose content was determined quantitatively by the amount of reducing power7 liberated after hydrolysis in 1 N hydrochloric acid a t 100" for 1 hour. These data are presented in Table 11. (6) R. L. Whistler and D . F. Durso, THISJOURNAL, 72, ti77 (1950). (7) N. Nelson, J . B i d . Chem., 153, 375 (1944).

T h e a-methylphenylhydrazone of galactose8 was prepared from the acid hydrolysates of A and B. Seventy-three mg. of fraction A was heated in a boiling water-bath for 1 hour with 1 N hydrochloric acid. The reaction mixture was then cooled and neutralized with a few drops of concentrated sodium hydroxide. Five ml. of reagent, containing 25 g. of a-methylphenylhydrazine and 3 ml. of glacial acetic acid in 100 ml. of absolute ethanol, was added to the neutralized reaction mixture. The solution was allowed t o stand a t room temperature for 13 hours and then a t 4' for 21 hours. Crystals began t o form about 1 hour after the reagent was ' the crystals were filtered on a added. After 24 hours a t 4 sintered glass funnel and washed with 3 ml. of cold absolute ethanol (yield 71.6 mg., m.p. and mixed m.p. 189-191'). Anal. Calcd. for ClaH?aNzOs: C, 54.90; H , 7.01; N, 9.87. Found: C,54.04; H, 7.06; N, 9.77. From 100mg. of crystalline fraction B, 122.6 mg. of the a-methylphenylhydrazone of D-galactose was obtained. When a control containing similar quantities of glycerol and galactose was heated with hydrochloric acid and carried through the same procedure, an 88% yield of a-methylphenylhydrazone of galactose was obtained. A chromatographic comparison of fraction A with a - ~ galactopyranosyl-2-glycerol revealed that the two substances were not identical (see Table 111). I n addition, fraction A had a specific rotation of +3.77' while a-n-galactopyranosyl-2-glycerol has a specific rotation of 4-166". Periodate Oxidation Studies.-A 0.1-millimole sample of the galactoside was placed in a 50". volumetric flask and water was added t o a volume of approximately 25 ml. Five ml. of 0.1 hl sodium periodate was added and the mixture diluted t o the 50-ml. mark. A blank containing the reagents but no galactoside was used. The solutions were allowed t o stand a t room temperature in the dark for 30 hours. The amount of periodate consumed in the reaction was estimated by removing 5-ml. samples a t various time intervals and treating with an excess of 0.05 M sodium arsenite in the presence of potassium iodide and sodium bicarbonate and back-titrating with 0.01 M iodine. For the samples were removed, determination of formic acid, 5". treated with 0.05 ml. of ethylene glycol and, after 5 minutes, titrated with 0.01 N sodium hydroxide using phenol red as an indicator. Formaldehyde was measured by its reaction with chromatropic acid according t o Speck and Foristg and also by precipitation with dimedon.la T h e results showed that the monogalactosylglycerol consumed 3.1 moles of periodate, giving rise to 1.0 mole of formic acid and 1.2 moles of formaldehyde. The digalactosylglycerol consumed 5.0 moles of periodate, giving rise to 1.9 moles of formic acid and 1.3 moles of formaldehyde. Enzymatic Studies.-In these studies a yeast hexokinase preparationll contaminated with a-galactosidase activity and a purified &galactosidase from E . coZP were used. T h e a-galactosidase studies were carried out in 0.05 M acetate buffer, p H 6.0 and the ,%galactosidase in 0.1 M potassium acid phosphate containing 0.01 M methionine. Each enzyme preparation was shown to be specific for its particular type of linkage using melibiose (a-galactoside) and lactose (@-galactoside)as test substrates. The a-galactosidase preparation was also shown t o hydrolyze a-D-galaCtOpy(8) E. L . Hirst, J. IC. N. Jones and E. A. Woods. .7. Chem. Soc., 1048 (1947). (9) J . C. Speck, Jr., and A . A. Forist, A n a l . Chem., 26, 1942 (1954). (10) B. H.Kicolet and L. .4. Shinn, f. Bioi. ('hem., 139,687 (19-11). (11) Kindly supplied by Dr. J . Lamer, Department of Chemistry, University of Illinois, Urbana, Ill. (12) Kindly supplied by Dr. S. Spiegelman, Department sf Bacteriology, University of Illinois, Urbana, Ill.

J. E. MILKSAND C. B. PURVES

3738

ranosyl-2-glycerol t o approximately 70% of the theoretical value in 2 hours at 30'. The enzyme and substrate, in the appropriate buffers, were incubated at 30" and at various time intervals aliquots were withdrawn and deproteinized with barium hydroxidezinc sulfate. Release of reducing power was determined by the Nelson8 procedure. KO hydrolysis of the digalactosylglycerol with the Bgalactosidase was observed. Incubation with a-galactosidase resulted in the release of only 50% of the potential reducing power. In order t o determine the products of this hydrolysis a larger scale experiment was run. Ten mg. of the digalactosylglycerol was incubated with @-galactosidase for 3 hours at 25". The entire reaction mixture was then deproteinized with barium hydroxide-zinc sulfate. The supernatant solution was put on a 3 X 6 cm. carbon-Celite

[CONTRIRT'TIOS FROM THE

VOl. 78

column (Darco G-60 anti Celite 545 iii a 1: 1 ratio). Tlic column was first washed with 1000 ml. of water and then with approximately 200 ml. of 10% ethanol. These two fractions were concentrated in vacuo and finally lyophilized. Paper chromatography of the material from the water wash revealed the presence of a reducing sugar corresponding to galactose. T h e material from the ethanol wash was nonreducing and was identical with the monogalactosylglycerol of fraction A as judged by paper chromatography. Paper Chromatography of an acid hydrolysate of this material revealed the presence of galactose and glycerol. The monogalactosj-lglycerol was not hydrolyzed by agalactosidase during 3 hours incubation at 30'. Incubation with the p-galactosidase resulted in a release of approximately 80% of the potential reducing power in 90 minutes. URBAXA, ILLINOIS

DIVISIOXOF ISDUSTRIAL A N D CELLULOSE CHEMISTRY, M C G I L L UNIVERSITTI., DlvIsIos, P U L P A S D PAPER RESEARCH INSTITLTE O F CANADA]

A N I ) THE \ v O O l >

CHEMISTRY

Attempted Preparation of a Homogeneous Hemicellulose from Aspen Wood BY J.

E.bIILKS AND

c. R.PURVES

RECEIVED DECEMBER 9, 1955 Chlorite holocellulose from aspen wood (Populus tremuloides) mas extracted with nearly anhydrous liquid ammonia near 20' and 150 p.s.i. The ammonia removed 10.6% of the wood in the form of acetamide, modified lignin and carbohydrate and rendered a further 9.6% soluble in subsequent extractions with water near 70". Crude pectins were eliminated from the latter extracts by bleaching with chlorite and b y acetylation followed by fractional precipitation. Deacetylation yielded about 1% of the wood as four fractions each containing one hexuronic and two 4-0-methylglucuronic acid residues combined with about 22, 16, 13 and 12 anhydro-D-xylose units, respectively. Partial hydrolysis yielded a crystalline aldotriuronic acid containing one 4-0-methylglucuronic acid and two xylose residues; also another (uncrystallized) in which one of the xylose residues was thought t o be replaced by t h a t of the unknown hexuronic acid. A chemical study by standard methods showed that the hemicellulose was based on anhydro-D-xylose units linked 1-4, probably with many branches in the second and third positions.

Although hemicelluloses extracted from aspen pulps by aqueous alkali have been examined on several occasion^,^-^ the literature records no sustained effort to separate a chemical individual from the original mixture. The present research had the isolation of such a product as one of its objectives, and another was to gain more information about the action of anhydrous, or nearly anhydrous, liquid ammonia on wood pulps. Liquid ammonia a t temperatures between 25 and 100' (about 150 to 900 p s i ) was known t o dissolve only 3.2 and 8.9% of solvent-extracted spruce and maple wood, respectively,8but a portion of the maple wood residue was rendered soluble in hot water.g The additional solubility was tentatively attributed to the cleavage of ester groups of an unknown type by the liquid ammonia, because the final wood residue contained 0.25y0 of additional nitrogen as amide and because at least 6670 of the acetyl groups in the wood was recovered as acetamide. The acetyl groups in birch and spruce holocelluloses, however, were unaffected when the liquid ammonia was used a t atmospheric (1) Abstracted from a Ph.D. Thesis submitted by J. E. Milks, October 1953. Presented before the Division of Cellulose Chemistry of the American Chemical Society, Minneapolis, September, 1956. (2) R. E. March, Tech. Assoc. Papers, 91, 240 (1948). (3) J . 0. Thompson and L. E. Wise, T a p p i , 5 5 , 331 (1952). (4) (a) J. K. N. Jones a n d L. E. Wise, J . Chem. Soc., 2750 (1962); (b) 3389 (1952). (5) J. Saarnio, I(. W a t h i n and C . Gustafsson, Paper and Timbev Jouynal (Finland), S6, 209 (1984). (6) D. C . Lea, T a p p i , 3 3 , 3 9 3 (1951). (7) M. A. Millett and A. J. S t a m m , J . P h y s . Colloid Chem., 51, 134 (19 l i ) . (8) hr. &I, >-an and C. €3. Purves, unpublished ( 9 1 I,. G. Neubauer, M. M Yan and C. B. Purves, uupublisbed

pressure and -33',loa or under less drastic conditions. An aspen holocellulose, rather than the wood itself, was used on the present occasion in order to facilitate the isolation of the hemicelluloses. Chlorite holocellulose" mas preferred to the product obtained by chlorination in water near 0°12313followed by extraction with alcohol-pyridine, l 4 because the former method left the holocellulose with a much lower nitrogen content (0.04 yo)than did the latter (0.27yG).I5 An increase in nitrogen content a t this point would complicate the interpretation of any change caused by the subsequent use of liquid ammonia. The chlorite holocellulose (Fig. 1) was then thoroughly extracted with hot water to remove material that might have contaminated the hemicelluloses to be extracted later. This material (extract-1) consisted of polysaccharides and of lignin oxidized in variable degree, but no evidence of a lignin-carbohydrate compound could be found. *kfter being dried, the residual holocellulose was extracted three times under pressure with anhydrous liquid ammonia, which removed lO.2y0of the wood (10) (a) K . J . Bjorkqvist a n d L. Jorgensen, Acta Chem. Scand., 6 . 978 (1951); (b) 6, 800 (1952). (11) L. E. Wise, M . Murphy and A. A. D'Addieco, Paper T r a d e J , 122, No. 2 , 3 5 (Jan. 10, 1946). (12) D. A. Sitch, Pulp and Paper M a g . C a n a d a , 60, 234 (Convention Issue, 1949). (13) T. E. Time11 and E. C. Jahn, Soensk Pappersfidn., 64, 831 (1951). The two methods were systematically reviewed a n d applied to paper birch. (14) G. J. Ritter and E. F. K u r t h , I n d . Eng. Chem., 26, 1250 (1933). (15) (a) B. B. Thomas, Pam' I n d . and P a p e v W o r l d , 26, 1281 (1945): ( h ) 27. 374 (1915), noted the increased nitrogen content of chlorincethanolamine holocelluloses